多链路即时通信中交互数据异常点检测仿真  

Simulation of Interactive Data Anomaly Detection in Multi-link Instant Communication

作　　者：赵磊[1] ZHAO Lei(Dongchang College,Liaocheng University,Liaocheng Shandong 252000,China)

机构地区：[1]聊城大学东昌学院

出　　处：《计算机仿真》2019年第11期445-448,共4页Computer Simulation

摘　　要：为了提高多链路即时通信网络的可靠性,需要对通信中交互数据异常点进行检测。针对传统的多链路即时交互数据异常点检测中,普遍存在着检测准确率较低、完成时间过长、成本消耗较大等问题,提出多链路即时通信中交互数据异常点检测方法。仅考虑时间序列变换特征,将K-均值聚类和粒子群优化算法相结合对交互异常数据进行聚类,取得聚类目标函数,利用粒子群对其进行求解,得到交互异常数据偏差函数,对多链路即时通信中的异常数据进行偏差统计,确定多链路即时通信中交互数据异常点,实现异常点的检测。实验结果表明,所提方法检测准确率较高、完成时间较短、成本消耗较低,具有一定的应用前景。In order to improve the reliability of multi-link instant network, it is necessary to detect abnormal points of interactive data in communication. Traditionally, the detection accuracy is low, completion time is long and cost is high. Therefore, a method to detect the abnormal points of interactive data in multi-link instant communication was proposed. Only considering the characteristics of time series transformation, we combined K-means clustering with particle swarm optimization algorithm to cluster the interactive abnormal data, so as to get the clustering objective function which was solved by particle swarm optimization. Then the interactive abnormal data deviation function was obtained. Moreover, the deviation statistics was performed on abnormal data in multi-link instant communication, and the abnormal point of interactive data in the multi-link instant communication was determined. Thus, the detection for the abnormal point was achieved. Following conclusion can be drawn from experimental results. The proposed method has higher detection accuracy, shorter completion time and lower cost, so it has certain application prospects.

关 键 词：多链路 即时通信 交互数据 异常点检测 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]